Comparison of vegetation index compositing scenarios: BRDF versus maximum VI approaches

Abstract

Satellite sensors, such as the AVHRR, SPOT and soon to be launched MODIS, MISR, VEGETATION and GLI acquire bidirectional reflectance data under different solar illumination angles. These systems will capture the strong anisotropic properties that vary with relative amounts and types of vegetation and soil within each pixel. Therefore, some knowledge of the bidirectional reflectance distribution function (BRDF) is a requirement for successful interpretation of directional reflectance data and vegetation indices, and derivation of land-cover-specific biophysical parameters. The objectives of this research were: (a) to parameterize empirical and semi-empirical BRDF models for different land cover types and MODIS spectral bands, (b) utilize the BRDF models to correct off-nadir measurements to nadir-equivalent values for vegetation index (VI) compositing and biophysical interpretation and (c) compare different vegetation index compositing scenarios. High spectral and spatial resolution bidirectional reflectance factor (BRF) measurements from the ASAS flown on the NASA C-130B aircraft were used for the analysis. Leaf area index (LAI) measurements were made concurrently at most of the study sites which included deciduous and coniferous forest, grassland and shrub savanna land covers. The normalized difference vegetation index (NDVI) and modified VI (MVI) were selected as classifiers in five different vegetation index composite scenarios: a maximum VI based on apparent reflectance data, a maximum VI based on at-surface reflectance data, a BRDF standardized VI, based on at-surface reflectances at nadir view angle, a BRDF normalized VI, based on at-surface reflectances at nadir view and nadir sun angles, a normalized bidirectional VI distribution function (BVIF). Nadir-equivalent VI accuracy and predictability were evaluated for all compositing scenarios using the measured nadir observations as a reference. The results of the analysis emphasize the importance of standardizing BRF for vegetation index compositing schemes and retrieval of biophysical parameters.